adfa, p. 1, 2011.
© Springer-Verlag Berlin Heidelberg 2011
Evaluation of English to Urdu Machine Translation
Vaishali Gupta, Nisheeth Joshi, Iti Mathur
Apaji Institute, Banasthali University, Rajasthan, India
vaishali.gupta77@gmail.com,
nisheeth.joshi@rediffmail.com,
mathur_iti@rediffmail.com
Abstract.
This paper is based on the Evaluation of English to Urdu Machine Translation.
Evaluation measures the quality characteristic of the Machine Translation out-
put and is based on two approaches: Human Evaluation and Automatic Evalua-
tion. In this paper, we are mainly concentrating over Human Evaluation.
Machine Translation is an emerging research area in which human beings play
a very crucial role. Since language is so vast and because of its diverse in na-
ture, the accuracy is not maintained. To maintain this accuracy, Human Evalua-
tion is taken as a base. Human Evaluation can be used with different parame-
ters to judge the quality of sentences.
Keywords: Human Evaluation, Adequacy, Fluency, Comprehensibility.
1 Introduction
We live in a trans-lingual society where people from different linguistic back-
grounds live together. It is practically impossible for an individual to speak all
this languages. In this context MT proves to be a savior. In MT, research work
totally depends on the result of its evaluation. Since Machine Translation is
an automated system therefore it is not necessary that the system will pro-
vide us the accurate result. To know the accuracy of the result, evaluation is
required. We can evaluate any languages by two ways: Human Evaluation
and Automatic Evaluation.
In Human Evaluation, we have a source language and target language. These
two languages are compared by an expert on the basis of some parameters.
Since human is a valuable resource therefore it is costly and time consuming.